List your company

Natural Language Inference on MSciNLI (ECE and BS)

2.08ECECaliDist (As.)
0.557610.833821.1131.3862Jun 4, 2026

Evaluation results

MethodLinks
CaliDist (As.)LLM=QWEN-3 8B2026.062.0822.76
CaliDist (Sa.)LLM=QWEN-3 8B2026.062.8822.55
CaliDist (Pr.)LLM=LLAMA-3.1 8B2026.064.9821.82
CaliDist (As.)LLM=LLAMA-3.1 8B2026.065.7521.94
CaliDist (Sa.)LLM=LLAMA-3.1 8B2026.066.2321.58
CaliDist (Pr.)LLM=QWEN-3 8B2026.066.5123.44
Temperature ScalingLLM=LLAMA-3.1 8B2026.0612.122.38
VanillaLLM=LLAMA-3.1 8B2026.0625.7628.3
EntropyLLM=LLAMA-3.1 8B2026.0628.0530.75
FSDLLM=LLAMA-3.1 8B2026.0629.0929.09
ConsistencyLLM=LLAMA-3.1 8B2026.0634.334.84
Temperature ScalingLLM=QWEN-3 8B2026.0635.4635.72
EntropyLLM=QWEN-3 8B2026.0638.2339.07
VanillaLLM=QWEN-3 8B2026.0638.638.6
FSDLLM=QWEN-3 8B2026.0638.8739.38
ConsistencyLLM=QWEN-3 8B2026.0640.1440.14